cryoelectron microscopy of a nucleating model bile in ... · cryoelectron microscopy of a...

16
Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,* David Q.-H. Wang, # Martin C. Carey, # and Donald M. Small* *Department of Biophysics, Boston University School of Medicine, Boston, Massachusetts 02118, and # Department of Medicine, Gastroenterology Division, Brigham and Women’s Hospital, Harvard Medical School and Harvard Digestive Disease Center, Boston, Massachusetts 02115 USA ABSTRACT Because gallstones form so frequently in human bile, pathophysiologically relevant supersaturated model biles are commonly employed to study cholesterol crystal formation. We used cryo-transmission electron microscopy, comple- mented by polarizing light microscopy, to investigate early stages of cholesterol nucleation in model bile. In the system studied, the proposed microscopic sequence involves the evolution of small unilamellar to multilamellar vesicles to lamellar liquid crystals and finally to cholesterol crystals. Small aliquots of a concentrated (total lipid concentration 29.2 g/dl) model bile containing 8.5% cholesterol, 22.9% egg yolk lecithin, and 68.6% taurocholate (all mole %) were vitrified at 2 min to 20 days after fourfold dilution to induce supersaturation. Mixed micelles together with a category of vesicles denoted primordial, small unilamellar vesicles of two distinct morphologies (sphere/ellipsoid and cylinder/arachoid), large unilamellar vesicles, multilamellar vesicles, and cholesterol monohydrate crystals were imaged. No evidence of aggregation/fusion of small unilamellar vesicles to form multilamellar vesicles was detected. Low numbers of multilamellar vesicles were present, some of which were sufficiently large to be identified as liquid crystals by polarizing light microscopy. Dimensions, surface areas, and volumes of spherical/ellipsoidal and cylindrical/arachoidal vesicles were quantified. Early stages in the separation of vesicles from micelles, referred to as primordial vesicles, were imaged 23–31 min after dilution. Observed structures such as enlarged micelles in primordial vesicle interiors, segments of bilayer, and faceted edges at primordial vesicle peripheries are probably early stages of small unilamellar vesicle assembly. A decrease in the mean surface area of spherical/ellipsoidal vesicles was correlated with the increased production of cholesterol crystals at 10 –20 days after supersaturation by dilution, supporting the role of small unilamellar vesicles as key players in cholesterol nucleation and as cholesterol donors to crystals. This is the first visualization of an intermediate structure that has been temporally linked to the development of small unilamellar vesicles in the separation of vesicles from micelles in a model bile and suggests a time-resolved system for further investigation. INTRODUCTION The structures and mechanisms involved in cholesterol sol- ubility and cholesterol crystal formation in human bile have been investigated for many years in the attempt to under- stand human gallstone formation (Cabral and Small, 1989; Carey and Small, 1978; Small, 1980; Wang and Carey, 1996a). Structural information about cholesterol carriers in pathophysiologically relevant model bile systems (Cohen et al., 1993; Gilat and So ¨mjen, 1996) as well as crystallization has been provided by a number of techniques such as quasielastic light scattering (Cohen et al., 1990; Little et al., 1991; Mazer et al., 1980; Mazer and Carey, 1983), small- angle x-ray scattering (So ¨mjen, 1994), small-angle neutron scattering (Hjelm et al., 1992; Long et al., 1994a), video- enhanced contrast microscopy (Halpern et al., 1986a), po- larizing light microscopy (Konikoff and Carey, 1994; Koni- koff et al., 1992; Wang and Carey, 1996a,b), and electron microscopy techniques such as negative staining (Groen et al., 1989; So ¨mjen et al., 1990b; Tao et al., 1993), freeze fracturing (So ¨mjen et al., 1986; van de Heijning et al., 1994), and vitreous ice cryomicroscopy (Kaplun et al., 1997, 1994; Vinson et al., 1989; Walter et al., 1991). The electron microscopy technique in vitreous ice (Dubochet et al., 1988) can provide images of particles and structures in lipid-rich systems that are believed to be close to their original state in solution. Utilizing this technique, we have attempted to detect the earliest stages of nucleation, visual- izing all structures present in a supersaturated, nucleating model bile from initial formation to 20 days after supersat- uration, and to correlate ultrastructural observations with those made with polarizing light microscopy. In the earliest stage, we have obtained images of the nucleation of micelles into primordial vesicles. MATERIALS AND METHODS Chemicals Grade A taurocholate (Sigma Chemical Co., St. Louis, MO) was purified by the method of Pope (1967) and found to be 99% pure by high- performance liquid chromatography (HPLC). Cholesterol (Nu-Chek Prep., Elysian, MN) and grade I egg yolk lecithin (Lipid Products, South Nutfield, Surrey, England) were found to be 99% pure by gas-liquid chromatog- raphy, thin-layer chromatography, and HPLC. All other chemicals and solvents were American Chemical Society (ACS) or reagent grade quality (Fisher Scientific Co., Medford, MA). Received for publication 9 March 1998 and in final form 4 November 1998. Address reprint requests to Mr. Donald L. Gantz, Department of Biophys- ics, W301, Boston University School of Medicine, 715 Albany Street, Boston, MA 02118. Tel: 617-638-4017; Fax: 617-638-4041; E-mail: [email protected]. © 1999 by the Biophysical Society 0006-3495/99/03/1436/16 $2.00 1436 Biophysical Journal Volume 76 March 1999 1436 –1451

Upload: others

Post on 01-Jun-2020

8 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice:Formation of Primordial Vesicles

Donald L. Gantz,* David Q.-H. Wang,# Martin C. Carey,# and Donald M. Small**Department of Biophysics, Boston University School of Medicine, Boston, Massachusetts 02118, and #Department of Medicine,Gastroenterology Division, Brigham and Women’s Hospital, Harvard Medical School and Harvard Digestive Disease Center, Boston,Massachusetts 02115 USA

ABSTRACT Because gallstones form so frequently in human bile, pathophysiologically relevant supersaturated model bilesare commonly employed to study cholesterol crystal formation. We used cryo-transmission electron microscopy, comple-mented by polarizing light microscopy, to investigate early stages of cholesterol nucleation in model bile. In the systemstudied, the proposed microscopic sequence involves the evolution of small unilamellar to multilamellar vesicles to lamellarliquid crystals and finally to cholesterol crystals. Small aliquots of a concentrated (total lipid concentration � 29.2 g/dl) modelbile containing 8.5% cholesterol, 22.9% egg yolk lecithin, and 68.6% taurocholate (all mole %) were vitrified at 2 min to 20days after fourfold dilution to induce supersaturation. Mixed micelles together with a category of vesicles denoted primordial,small unilamellar vesicles of two distinct morphologies (sphere/ellipsoid and cylinder/arachoid), large unilamellar vesicles,multilamellar vesicles, and cholesterol monohydrate crystals were imaged. No evidence of aggregation/fusion of smallunilamellar vesicles to form multilamellar vesicles was detected. Low numbers of multilamellar vesicles were present, someof which were sufficiently large to be identified as liquid crystals by polarizing light microscopy. Dimensions, surface areas,and volumes of spherical/ellipsoidal and cylindrical/arachoidal vesicles were quantified. Early stages in the separation ofvesicles from micelles, referred to as primordial vesicles, were imaged 23–31 min after dilution. Observed structures such asenlarged micelles in primordial vesicle interiors, segments of bilayer, and faceted edges at primordial vesicle peripheries areprobably early stages of small unilamellar vesicle assembly. A decrease in the mean surface area of spherical/ellipsoidalvesicles was correlated with the increased production of cholesterol crystals at 10–20 days after supersaturation by dilution,supporting the role of small unilamellar vesicles as key players in cholesterol nucleation and as cholesterol donors to crystals.This is the first visualization of an intermediate structure that has been temporally linked to the development of smallunilamellar vesicles in the separation of vesicles from micelles in a model bile and suggests a time-resolved system for furtherinvestigation.

INTRODUCTION

The structures and mechanisms involved in cholesterol sol-ubility and cholesterol crystal formation in human bile havebeen investigated for many years in the attempt to under-stand human gallstone formation (Cabral and Small, 1989;Carey and Small, 1978; Small, 1980; Wang and Carey,1996a). Structural information about cholesterol carriers inpathophysiologically relevant model bile systems (Cohen etal., 1993; Gilat and Somjen, 1996) as well as crystallizationhas been provided by a number of techniques such asquasielastic light scattering (Cohen et al., 1990; Little et al.,1991; Mazer et al., 1980; Mazer and Carey, 1983), small-angle x-ray scattering (Somjen, 1994), small-angle neutronscattering (Hjelm et al., 1992; Long et al., 1994a), video-enhanced contrast microscopy (Halpern et al., 1986a), po-larizing light microscopy (Konikoff and Carey, 1994; Koni-koff et al., 1992; Wang and Carey, 1996a,b), and electronmicroscopy techniques such as negative staining (Groen etal., 1989; Somjen et al., 1990b; Tao et al., 1993), freeze

fracturing (Somjen et al., 1986; van de Heijning et al.,1994), and vitreous ice cryomicroscopy (Kaplun et al.,1997, 1994; Vinson et al., 1989; Walter et al., 1991). Theelectron microscopy technique in vitreous ice (Dubochet etal., 1988) can provide images of particles and structures inlipid-rich systems that are believed to be close to theiroriginal state in solution. Utilizing this technique, we haveattempted to detect the earliest stages of nucleation, visual-izing all structures present in a supersaturated, nucleatingmodel bile from initial formation to 20 days after supersat-uration, and to correlate ultrastructural observations withthose made with polarizing light microscopy. In the earlieststage, we have obtained images of the nucleation of micellesinto primordial vesicles.

MATERIALS AND METHODS

Chemicals

Grade A taurocholate (Sigma Chemical Co., St. Louis, MO) was purifiedby the method of Pope (1967) and found to be �99% pure by high-performance liquid chromatography (HPLC). Cholesterol (Nu-Chek Prep.,Elysian, MN) and grade I egg yolk lecithin (Lipid Products, South Nutfield,Surrey, England) were found to be �99% pure by gas-liquid chromatog-raphy, thin-layer chromatography, and HPLC. All other chemicals andsolvents were American Chemical Society (ACS) or reagent grade quality(Fisher Scientific Co., Medford, MA).

Received for publication 9 March 1998 and in final form 4 November 1998.

Address reprint requests to Mr. Donald L. Gantz, Department of Biophys-ics, W301, Boston University School of Medicine, 715 Albany Street,Boston, MA 02118. Tel: 617-638-4017; Fax: 617-638-4041; E-mail:[email protected].

© 1999 by the Biophysical Society

0006-3495/99/03/1436/16 $2.00

1436 Biophysical Journal Volume 76 March 1999 1436–1451

Page 2: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

Preparation of model bile

Model micellar bile was prepared by coprecipitation of cholesterol, leci-thin, and taurocholate from CHCl3-MeOH (2:1, v/v) to yield a relativecomposition of 8.5%, 22.9%, and 68.6%, respectively (all mole %). Thetotal lipid concentration was 29.2 g/dl. Lipid mixtures were dried under astream of N2 and then under reduced pressure for 24 h to achieve constantweight. Each dried lipid film was then dissolved in aqueous solution (0.15M NaCl adjusted with 1–2 �l of 1 M NaOH to pH 7.0), containing 3 mMNaN3 as an antimicrobial agent. After the tubes were sealed under a blanketof N2 with Teflon-lined screw caps, they were vigorously vortex-mixed fora few minutes. To ensure complete solubilization of coprecipitated cho-lesterol in micelles, the model bile was shaken at 100 rpm for 1 h at 50°C,using a model 75 Wrist Action Shaker (Burrell Corporation, Pittsburgh,PA). The stock bile solution was then microfiltered at 22°C through apreheated Swinnex-GS filter assembly containing a 0.22-�m filter (Milli-pore Products Division, Bedford, MA), and �0.1 ml of sample wasaspirated for later lipid analysis (see below). To induce supersaturation,stock solutions were diluted fourfold with 0.15 M NaCl (pH 7.0) to 7.3g/dl. According to critical tables (Carey, 1978), with the assumption thatthe intended cholesterol composition was correct (see Wang and Carey,1996b), this dilution should increase the cholesterol saturation index (CSI)from 0.97 � 0.03 to 1.21 � 0.05 (see Lipid Analysis below). The dilutiontime point was taken as the initiation of all nucleation studies (Wang andCarey, 1996a). Fig. 1 indicates that this model bile at equilibrium falls ina central three-phase zone, region C, containing saturated micelles, liquidcrystals, and cholesterol monohydrate crystals, with a microscopic crystal-lization sequence of micelles3 liquid crystals3 cholesterol monohydratecrystals 3 anhydrous cholesterol crystals, which is well defined (Wangand Carey, 1996a).

Monitoring of crystallization sequences bypolarizing light microscopy

The nucleating biles were sealed in glass test tubes, blanketed with N2, andincubated at room temperature (22 � 1°C). Crystallization sequences were

verified by the method of Wang and Carey (1996a). No differences incrystallization sequences were observed if the sample was shaken beforesampling or left undisturbed. In brief, 5-�l samples were taken from thebottom third of a 4-ml model bile sample and placed immediately on aglass slide at room temperature (22 � 1°C). Samples were observedinitially without a coverslip by polarized light microscopy (Photomicro-scope III; Carl Zeiss, Thornwood, NY) at 400� magnification. The samplewas then compressed with a cover glass and reexamined with phase-contrast optics. New slides were examined at 2-h intervals for 12 h, andthen daily for 20 days to determine crystallization sequences. Detectiontimes of cholesterol crystals and liquid crystals were defined as the timefrom initial dilution to the earliest microscopic detection at 400� of arc,helical, tubular, cholesterol monohydrate, or small, aggregated, and fusedliquid crystals (Wang and Carey, 1996a).

Lipid analysis

Taurocholate concentrations were assayed by the 3�-hydroxysteroid dehy-drogenase method (Turley and Dietschy, 1978); lecithin by an inorganicphosphorus procedure (Bartlett, 1959); and cholesterol enzymatically(Fromm et al., 1980). Calculation of CSIs for model bile was based on thecritical tables (Carey, 1978).

Cryo-transmission electron microscopy

Model bile was incubated at 22°C in a 15 � 100 mm screw cap glassculture tube (Kimble/Kontes, Vineland, NJ) on a bench top during eachexperiment (20 days). The model bile initially extended to a height of 3 cmin the tube. The tube was hand-shaken for �30 s just before sampling orwas left undisturbed. Each 7-�l aliquot was removed with a Gilson P20pipetteman (Villiers-le-Bel, France) at a height of 0.8 cm from the tube’sbase and placed on a freshly glow-discharged (Dubochet et al., 1982) holeycarbon film covering a 400 mesh copper grid (Electron Microscopy Sci-ences, Fort Washington, PA), which was gripped by forceps held closed bya rubber ring. The forceps were then attached to a nitrogen gas-drivenplunger (Charles Ingersoll Corp., Winthrop, MA) inside a plexiglas freez-ing station (Dan-Kar Corp., Reading, MA) of dimensions 21� � 21� � 31�(height) with a vertically sliding access door of dimensions 15� � 9�(height). This door was kept closed except when a sample was loaded, agrid was blotted, or a grid was transferred to a storage box. Humid airproduced by an Ultrasonic Humidifier (model TUH-420; Tatung Companyof America, Long Beach, CA) entered at the rear of the freezing stationthrough a 1.5�-diameter flexible hose. The atmosphere in the chamberbecame foggy, particularly over the container of liquid nitrogen where theforceps were hung. Humidity was monitored with a Vaisala HumidityMeter (Vaisala Co., Helsinki, Finland). The humidity dipped slightly below100% when the access door was opened during blotting. Samples wereblotted from the underside (not the sample side) with oven-dried no. 1 filterpaper (Whatman Paper Limited, Kent, England) and plunged, by activatinga pedal, into a liquid ethane-filled stainless steel cup partially submerged inliquid nitrogen. Because the temperature of liquid nitrogen (77 K) is belowthe freezing point of ethane (90 K), the ethane slowly solidified from theinside surface of the cup toward the center. Each grid was plunged after athin layer of solid ethane had appeared, ensuring that the remainder of theliquid ethane was close to its freezing point. Underside blotting (Toyo-shima, 1989) was used to concentrate particles. Pilot experiments indicatedthat 1–10 s between blotting and plunging in the humidified environmentwould provide suitable vitreous ice thicknesses. During the first 30 minafter supersaturation, grids were prepared as quickly as possible, resultingin the time points shown in Table 1. At time points of 6 h and 1, 4, 10, and20 days, several grids were prepared with postblotting/preplunging times of1, 5, and 10 s. Estimates of cooling times required to produce vitreous icefrom thin liquid films such as ours range from 2 ms (Klosgen and Helfrich,1993) to an optimum of 0.1 ms (Dubochet et al., 1988). Vitrified sampleswere stored in a cryogenic storage dewar (Taylor-Wharton, Theodore, AL)under liquid nitrogen.

FIGURE 1 Equilibrium phase diagram of taurocholate-lecithin-choles-terol system (total lipid concentration � 7.3 g/dl, 0.15 M NaCl, pH 7.0,22°C) used in this study. Composition was 68.6% taurocholate, 22.9%lecithin, and 8.5% cholesterol (mole %), placing it in region C of thecentral three-phase zone (F). The three phases at equilibrium are saturatedmicelles, liquid crystals, and cholesterol monohydrate crystals (see Fig. 2for crystallization sequences).

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1437

Page 3: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

Vitrified samples were transferred into a Single Tilt Cryoholder (model626), using a Cryotransfer Station (Gatan, Pleasanton, CA). The cryoholdermaintained samples at 99 K while images were recorded at a defocus valueof 1.4 � on SO-163 electron image film (Eastman Kodak Co., Rochester,NY) in a Philips CM-12 transmission electron microscope (Philips Elec-tron Optics, Eindhoven, The Netherlands) equipped with a low-dose kit tominimize irradiational damage. Edges and centers of large holes and entiresmall holes were photographed. The scope was calibrated with a siliconmonoxide grating replica (Ernest F. Fullam, Latham, NY). Micrographswere developed in full-strength D19 (Eastman Kodak Co.) for 12 min.Images of vesicles at 0° and 45° tilt were compared to confirm three-dimensional morphology.

Particle measurements were made with a Peak 7� magnifier with a1–20-mm graticule in 0.1-mm divisions on micrographs enlarged 2.5� onresin-coated paper. At each time point, particles were analyzed from atleast 18 vitreous ice-containing holes to a maximum of 81 holes. Ice at holeedges was thicker than ice in hole centers, but among all holes inspected nodata were obscured.

For negative staining a 30-�l aliquot of model bile sampled in the samemanner as vitreous ice was mixed with 10 �l of 2% osmium tetroxide in0.1 M phosphate buffer (pH 7.3), giving a final OsO4 concentration of0.5%. After a 20-min fixation to stabilize lipid structures, a 10-�l aliquotwas placed on a glow-discharged (Dubochet et al., 1982) carbon, formvar-coated 300 mesh copper grid, stained with 1% sodium phosphotungstate(1% NaPT, pH 7.3), blotted, and air-dried.

Classification of vesicle morphology

Vesicle-like particles that had curved or straight single strands and/or anoccasional bilayer fragment that formed a discontinuous circular to ovaloutline sufficient for diameter measurement were considered a new cate-gory of vesicle. These were called “primordial vesicles.” Small unilamellarvesicles were completely formed and had distinct, well-defined continuousbilayers. Each small unilamellar vesicle was assigned to one of twomorphological categories based on its two-dimensional projection: spher-ical/ellipsoidal vesicle or cylindrical/arachoidal vesicle. The morphologiesof small unilamellar vesicles were confirmed by 45° tilt images. A vesiclewith a circular or oval-shaped projection was a sphere or ellipsoid, respec-tively. A vesicle with a projection of constant width along its entire lengthwas a cylinder. A vesicle having a peanut-shaped projection with at least

one end of greater width than its center was an arachoid. A number ofparameters determining vesicle shape have been discussed in detail by Muiet al. (1995).

Calculation of surface area and volume of smallunilamellar vesicles

The following formulae (Frame, 1992) were used to calculate the surfacearea and volume of small unilamellar vesicles:

Surface area of spherical vesicle � 4�r2

Surface area of ellipsoidal vesicle � 2�(r1)2 � 2�(r2r1 e)arcsin(e),where r1 is the minor radius, r2 is the major radius, and e (eccentricity) �the square root of [1 (r1)2 (r2)2].

Surface area of cylindrical vesicle � �dL � 2�r2, where d is thediameter of the vesicle and L is the length of the vesicle.

Surface area of arachoidal vesicle � �d(avg)L � 2�r2(avg), whered(avg) � [d(min) � 2d(max)] 3, d(min) is the smallest width at thecenter of the vesicle, d(max) is the mean of the largest width at each endof the vesicle, and r(avg) � d(avg) 2.

Volume of spherical/ellipsoidal vesicles � (4/3)�(r12).

Volume of cylindrical/arachoidal vesicles � �r2(avg)L.

Statistical methods

All data are expressed as means � SD. Statistical methods, includingparametric and nonparametric tests, �2, and least-squares regression anal-yses were performed with an RS1 software package, release 4.3.1, 1991(BBN Software Products Corporation, Cambridge, MA). Statistical signif-icance was defined as a two-tailed probability less than 0.05.

RESULTS

Crystallization sequences by polarizinglight microscopy

The time sequences of the appearance and disappearance ofliquid crystals, cholesterol monohydrate crystals, arcs, he-lices, and tubular crystals in this model bile system asobserved by light microscopy are shown in Fig. 2 A. Eachtype of crystal is illustrated in Fig. 2 B. Small, aggregated,and fused liquid crystals (Fig. 2 B: 1, 2, 3) were detectedfirst at 1, 2, and 3 days, respectively, after fourfold dilutionto induce supersaturation. They peaked at 6, 7, and 9 daysand disappeared at 14, 15, and 16 days, respectively. Cho-lesterol monohydrate crystals (Fig. 2 B: 4) were first ob-served at 4 days and peaked at 17 days. Arc, helical, andtubular crystals (Fig. 2 B: 5, 6, 7, 8) were initially detectedat 11, 12, and 13 days and peaked at 16, 17, and 18 days,respectively. Detailed information on crystallization path-ways in model biles of similar composition has been pub-lished (Wang and Carey, 1996a). No appreciable differ-ences were observed in the crystallization sequences whensamples were taken from an unshaken tube of model bile orfrom a tube shaken just before bile was sampled.

Particles observed by cryo-transmissionelectron microscopy

Six types of nucleating particles and cholesterol monohy-drate crystals were visualized by vitreous ice electron mi-

TABLE 1 Particles and crystals observed in nucleatingmodel bile by cryo-TEM at various times aftersupersaturation*

Time Micelles PVs#

SUVs

LUVs MLVsChM

crystalsSEVs CAVs

2 min � 5 min � 9 min � 13 min � 23 min � � 27 min � � �§ 31 min � � 6 h � � � � 1 day � � � � � 4 days � � � � � � 10 days � � � � �20 days � � � � � �

*Sample tube was shaken prior to sampling; (�), present; (), absent.#PVs, primordial vesicles; SUVs, small unilamellar vesicles; SEVs, spher-ical/ellipsoidal vesicles; CAVs, cylindrical/arachoidal vesicles; LUVs,large unilamellar vesicles; MLVs, multilamellar vesicles; ChM, cholesterolmonohydrate.§Only one CAV was seen in 15 fields.

1438 Biophysical Journal Volume 76 March 1999

Page 4: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

croscopy: 1) micelles—mostly ovoid/barrel-shaped parti-cles of mean width and length (nm � SD) of 3.6 � 1.1 and5.6 � 1.6, respectively (n � 98); 2) primordial vesicles—an

early stage in the formation of spherical/ellipsoidal vesiclesin which a discontinuous circular or oval outline apparentlyformed by short segments of bilayer fragments is visible; 3)

FIGURE 2 Based on the principalcharacteristics of the phases for theabove model bile, its crystallization se-quence was in crystallization pathway C(Wang and Carey, 1996a). (A) The ver-tical axis represents arbitrary numbersof crystals and liquid crystals per micro-scopic field at �400 magnification, allnormalized to the same maximum. Con-secutive numbers represent small (1),aggregated (2), and fused (3) liquidcrystals, plate-like cholesterol monohy-drate crystals (4), and arc-shaped (5),helical (6), and tubular (7) crystals. Thearrow indicates the first detection ofcholesterol monohydrate crystals. (B)Characteristics of liquid and solid crys-tals of cholesterol observed by polariz-ing light microscopy in this study: (1)small nonbirefringent liquid crystals; (2)nonbirefringent aggregated liquid crys-tals; (3) typical fused liquid crystalswith Maltese cross birefringence and fo-cal conic textures; (4) typical choles-terol monohydrate crystals, with 79.2°and 100.8° angles, and frequent notchedcorners; (5) arc-like (possibly anyhy-drous cholesterol) crystal; (6) regularright-handed helical crystals; (7) tubularcrystal; and (8) tubular crystal fracturingat the ends to produce plate-like choles-terol monohydrate crystal. All magnifi-cations are �800. Bar � 10 �m.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1439

Page 5: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

small unilamellar vesicles—two morphologically uniquecategories of small unilamellar vesicles, spherical/ellipsoi-dal vesicles, and cylindrical/arachoidal vesicles, with dis-tinct bilayers and mean diameters less than 100 nm; 4) largeunilamellar vesicles—vesicles with distinct bilayers andmean diameters greater than 100 nm; 5) multilamellar ves-icles—vesicles with more than one bilayer; and 6) choles-terol monohydrate crystals—large platelike crystals withunique edge angles of 79.2° and 100.8°. Attempts at obtain-ing images of undiluted model bile at �30 g/dl were un-successful because vitreous ice layers were too dense. Noparticles were seen in buffer (0.15 M NaCl, pH 7.0) alone(micrograph not shown).

The dispersed particles and crystals observed at varioustime points after supersaturation are enumerated in Table 1.In this experiment, the sample tube was shaken for 30 s justbefore removal of an aliquot for vitrification.

Micelles were present at all time points and are illustratedin Fig. 3 A. Ranges of width and length were 1.1–5.7 nmand 2.3–10.3 nm, respectively (n � 98). Circular regions ofreduced micelle granularity (Fig. 3 A, between arrows) mayrepresent the beginning stage of vesicle formation. Athigher magnification (Fig. 3 A, inset) the morphology of thenucleating micelles appeared rectangular with some squares

(rare rounded ends), suggesting barrel-shaped particles andnot discs.

Primordial vesicles were first detected 23 min after su-persaturation (Table 1, Fig. 3 B). Primordial vesicles weregenerally round to slightly oval in shape. Their interiorsappeared slightly more dense but less granular than sur-rounding ice with only micelles (a projection through twopartially formed membranes would likely increase density)and contained angular structures (with rare exceptions) ofmean width and length (nm � SD) of 4.4 � 1.8 and 6.2 �1.9, respectively (n � 34), a size similar to micelles (Fig. 3B). The width and length ranges of these angular structureswere 1.1–8.0 nm and 3.4–10.3 nm, respectively. At 27 min(Fig. 3 C), several of the primordial vesicles had boundariespartially defined by short segments of bilayer (ranging from2.3 to 4.6 nm in thickness) and by curved or straight singlestrands. These strands are probably not monolayers becausemonolayers would be very unstable and are not likely to beimaged. The end-to-end contact of several straight segmentsat obtuse angles gave a faceted appearance to some of theboundaries (Fig. 3 C). Primordial vesicles were also ob-served at 6 h and 4 days (rare), but not at 1, 10, and 20 daysafter supersaturation by dilution (Table 1). The mean widthand length (nm � SD) of primordial vesicles at 23 min was

FIGURE 3 Here and next two pages. Cryo-transmission electron microscopy micrographs of mixed micelles, primordial vesicles, spherical/ellipsoidalvesicles, and cylindrical/arachoidal vesicles observed in model bile vitrified at various time points after supersaturation by dilution (see Table 1). Micelleswere present at all time points. See Tables 2 and 3 for mean sizes of primordial vesicles and small unilamellar vesicles, respectively. Regions of higherdensity near hole edges indicate thicker ice. In A–E, bars � 50 nm. (A) At 2 min, mixed micelles of width and length (mean � SD) of 3.6 � 1.1 and 5.6 �1.6 nm were clearly seen at the hole periphery. Thin, curved threadlike structures (arrowheads) were visible toward the hole center. Circular areas ofreduced micelle granularity (between arrows) may be a very early stage in the vesicle development. (Inset) Micelles at higher magnification in a hole interiorat 2 min appeared square to rectangular (arrowheads), suggesting a barrel-shaped morphology. Bar � 25 nm.

1440 Biophysical Journal Volume 76 March 1999

Page 6: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

FIGURE 3 (Continued). (B) At 23 min, primordial vesicles (arrows) were present and contained angular structures that were slightly larger than micelles(see interiors). Also visible was an extremely rare cluster of lamellae (arrowhead). (C) At 27 min, primordial vesicles had slightly more well-definedboundaries, some of which appeared faceted (large arrowheads). Occasional short bilayer segments at vesicle peripheries were visible (small arrowheads).One cylindrical/arachoidal vesicle (arrow) was detected at this time point. No primordial cylindrical/arachoidal vesicles were ever seen.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1441

Page 7: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

FIGURE 3 (Continued). (D) At 6 h, spherical/ellipsoidal vesicles (small arrowheads) and cylindrical/arachoidal vesicles (large arrowheads) withwell-defined bilayers were prevalent. Primordial vesicles (arrows) and micelles were interspersed. (E) Small unilamellar vesicles were frequently observedin ranks arranged concentrically within a hole (spherical/ellipsoidal vesicles, small arrowhead; cylindrical/arachoidal vesicle, large arrowhead). Thinnerice is located toward the top.

1442 Biophysical Journal Volume 76 March 1999

Page 8: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

47 � 7.4 and 52 � 8.6, respectively (Table 2). There was nosignificant increase in the mean size of primordial vesiclesat 27 and 31 min, but primordial vesicles at 6 h and 4 dayswere significantly larger. Although a few round primordialvesicles were seen at each of the above time points, themean shape was oval, because the mean width was signif-icantly different from the mean length at all time points(Table 2).

Although two groups of bilayer-like structures were ob-served in our study, they were extremely rare (only two inover 450 photographed holes). A group of four such struc-tures detected at 23 min (Fig. 3 B) had a range of thick-nesses of 4.6–5.7 nm and lengths of 40–45 nm. The othergroup, observed at 10 days (not shown), contained struc-tures with a thickness of 6 nm and a range of lengths of57–97 nm.

Although the exact time point when small unilamellarvesicles first appeared was not determined, at 6 h smallunilamellar vesicles with distinct bilayers were common(Fig. 3 D). These vesicles could be divided into two mor-phologically distinct categories: spherical/ellipsoidal vesi-cles and cylindrical/arachoidal vesicles (see Materials andMethods for details). Three-dimensional morphologies wereconfirmed by 45° tilt images (not shown). Bilayer thick-nesses of spherical/ellipsoidal vesicles and cylindrical/ara-choidal vesicles ranged from 3.4 to 5.7 nm. Both vesiclemorphologies were commonly found at each of four subse-quent time points (Table 1). Some primordial vesicles wereinterspersed among spherical/ellipsoidal vesicles and cylin-drical/arachoidal vesicles (Fig. 3 D). Spherical/ellipsoidalvesicles and cylindrical/arachoidal vesicles were frequentlyin rows or ranks positioned in concentric rings from thinnerto thicker ice within a hole (Fig. 3 E). One arachoid vesiclewas observed at 27 min (Fig. 3 C), the only arachoiddetected during the first 31 min after supersaturation bydilution among 120 photographed holes.

Large unilamellar vesicles and multilamellar vesicleswere first detected at 1 day after supersaturation (Table 1)and increased slightly in number up to the final time pointof 20 days (data not shown). These vesicles were mostcommonly seen at hole edges where ice was thicker (Fig. 4,A–C). Multilamellar vesicles of spherical/ellipsoidal andcylindrical/arachoidal morphologies were present at eachtime point but not in sufficient numbers to analyze relative

frequencies. Diameters of observed large unilamellar vesi-cles ranged from 140 to 500 nm; those of multilamellarvesicles ranged from 55 nm to over 1 �m. The mean bilayerthickness of large unilamellar vesicles was 5.8 nm (Fig. 4B); that of multilamellar vesicles was 6.4 nm (Fig. 4 C).Multilamellar vesicles were occasionally seen inside largeunilamellar vesicles (Fig. 4 B). Multilamellar vesicles ob-served by cryo-transmission electron microscopy at 1 daysuch as that shown in Fig. 4 A were not large enough to bevisible by polarizing light microscopy (light microscopycould detect crystals as small as 500 nm). Larger multila-mellar vesicles such as one with dimensions of 640 nm �850 nm (Fig. 4 C) are potentially detectable by polarizinglight microscopy (Wang and Carey, 1996a; see Fig. 2).

Large crystals were observed at 10 and 20 days afterdilution (Fig. 4 D). The edge angles of these thin platelikecrystals were 79.2° and 100.8°, like those reported forcholesterol monohydrate (Small, 1986). The dimensions ofthe plate in Fig. 4 D are 8 �m � 21 �m. Frequently, theseplates appeared to be stacked upon one another. A thin,partially coiled structure appeared to be either superimposedupon the image of the plate or associated with the plate. Thiscoiled structure might be similar to arclike or helical struc-tures seen by others using light microscopy and cryo-trans-mission electron microscopy (Kaplun et al., 1994; Konikoffet al., 1992; Wang and Carey, 1996a), or could be thegrowth face of the crystal with a growth spiral starting in ascrew dislocation (Toor et al., 1978).

Another experiment using a model bile with the samecomposition but with samples taken from an undisturbedtube yielded qualitatively similar results in vitreous ice (datanot shown), with the exception that large unilamellar vesi-cles and cholesterol monohydrate crystals were not ob-served (compare Table 1). Failure to see large unilamellarvesicles and cholesterol monohydrate crystals may be at-tributable to heterogeneous distribution in the undisturbedsample tube (cholesterol density of 1.056 g/ml promotessettling; Konikoff and Carey, 1994) and to the low proba-bility of capture in thin vitreous ice layers.

Negatively stained preparations of model bile sampled at2 and 10 days after supersaturation produced structurescompletely different from those seen in vitreous ice (Fig. 5).These lamellar-like structures were stacks of straight toslightly curved discs or sheets of bilayer. Bilayer thick-nesses ranged from 5.0 to 5.3 nm. Widths of individual discsin stacks ranged from 16 to 117 nm.

Evolution of spherical/ellipsoidal vesicles andcylindrical/arachoidal vesicles

To aid in the investigation of the evolution and distinctive-ness of spherical/ellipsoidal vesicles and cylindrical/arach-oidal vesicles and in the evaluation of the proposed role ofthese vesicles as cholesterol carriers and donors, we calcu-lated their mean size, surface area, and volume at each offive time points: 6 h and 1, 4, 10, and 20 days (Table 3). The

TABLE 2 Mean dimensions of primordial vesicles

Time n Width* Length*

23 min 32 47.1 � 7.4 52.7 � 8.6#

27 min 52 48.6 � 6.3 53.1 � 6.6#

31 min 152 48.8 � 9.4 54.8 � 9.5#

6 h 56 74.0 � 8.9§ 78.7 � 8.5#§¶

4 days 9 80.2 � 8.3§ 86.0 � 6.8#§

*Data are expressed as mean nm � SD.#p � 0.005 compared to vesicle width.§p � 0.01 compared to 23, 27, and 31 min.¶p � 0.05 compared to 4 days.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1443

Page 9: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

mean widths of spherical/ellipsoidal vesicles at 6 h and 4,10, and 20 days and mean lengths at 6 h and 1, 4, 10, and20 days (Table 3) were signficantly larger (p � 0.0005) thanthose of primordial vesicles at 23, 27, and 31 min (Table 2).Although some spherical/ellipsoidal vesicles were spheri-cal, the mean shape of spherical/ellipsoidal vesicles wasellipsoidal because mean widths and lengths at all timepoints were significantly different (Table 3). The meanwidth, surface area, and volume of spherical/ellipsoidalvesicles increased significantly from 1 to 10 days, thendecreased significantly at 20 days. The mean length ofspherical/ellipsoidal vesicles increased significantly at 10

days and decreased at 20 days. Within the population ofcylindrical/arachoidal vesicles, mean surface area and vol-ume showed a downward trend from 6 h to 10 days.

There was a clear difference between the mean widthsand lengths of spherical/ellipsoidal vesicles and cylindrical/arachoidal vesicles (Table 3). In general, the mean surfaceareas of cylindrical/arachoidal vesicles were larger thanthose of spherical/ellipsoidal vesicles, and the mean vol-umes were smaller. The mean surface area and volume ofthe entire vesicle population were monitored over time bycombining and averaging the surface areas and volumes ofall spherical/ellipsoidal vesicles and cylindrical/arachoidal

FIGURE 4 Cryo-transmission electron microscopy micrographs of vitreous ice-embedded large unilamellar vesicles, multilamellar vesicles, andcholesterol monohydrate crystals formed in nucleating model bile at various time points after supersaturation by dilution. (A) Multilamellar vesicles (largearrowhead) observed at 24 h have dimensions of 50 � 140 nm. Cylindrical/arachoidal vesicles and micelles are also visible. Bar � 50 nm. (B) Largeunilamellar vesicles, shown here at 20 days, had diameters ranging from 140 to 500 nm and may be enlarged because of squeezing/flattening in the vitreousice layer. Note the multilamellar vesicle inside large unilamellar vesicles (arrowhead). The mean bilayer thickness was 5.8 nm. Bar � 100 nm. (C) Thislarge multilamellar vesicle seen at 20 days may be of sufficient size (0.64 �m � 0.85 �m) to be identified as a liquid crystal by light microscopy. However,artifactual enlargement is likely because of flattening in thin vitreous ice layers. The mean bilayer thickness was 6.4 nm. Bar � 100 nm. (D) A cholesterolmonohydrate crystal seen at 20 days had a 8-�m width and characteristic edge angles of 79.2° and 100.8°. Note looped structure (large arrowhead), largeunilamellar vesicles (small arrowheads), multilamellar vesicle (small arrow), and a large cylindrical/arachoidal vesicle (large arrow). Bar � 1 �m.

1444 Biophysical Journal Volume 76 March 1999

Page 10: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

vesicles at each time point (Table 4). A significant increasein surface area was noted 10 days after supersaturation, anda significant decrease at 20 days. A surface area/volumeratio (Table 4) of 1:9.0 indicates high volume relative tosurface area with a preponderance of spherical/ellipsoidalvesicles, whereas with more cylindrical/arachoidal vesiclespresent, a ratio of 1:7.3 indicates reduced volume relative tosurface area.

The evolution of primordial vesicles, spherical/ellipsoidalvesicles, and cylindrical/arachoidal vesicles can be seenmore clearly if the width of each particle is plotted versusthe ratio of its width to its length (Fig. 6). Spherical/ellipsoidal vesicles appear at the top of each graph (width/length ratios near 1.0), and elongated cylindrical/arachoidalvesicles near the bottom (ratios � 0.2–0.3). Two particleswith the same dimensions were superimposed. The distri-butions of primordial vesicles at 23 and 27 min were similar(Fig. 6, A and B), as were their mean widths and lengths(Table 2). A portion of the primordial vesicle population at31 min (Fig. 6 C) had diameters 1.5 times the populationaverage. At 6 h (Fig. 6 D), the mean diameter of thepopulation of primordial vesicles had increased signifi-cantly to a range of 60–90 nm (Table 2). Apparently, thatportion of the primordial vesicle population at 31 min thathad widths of �35–55 nm became spherical/ellipsoidalvesicles. No primordial cylindrical/arachoidal vesicles were

ever seen. The populations of spherical/ellipsoidal vesiclesand cylindrical/arachoidal vesicles each became more ho-mogeneous: the population of spherical/ellipsoidal vesiclesevolving toward spheres (Fig. 6, F and H), and the popula-tion of cylindrical/arachoidal vesicles toward longer parti-cles (Fig. 6, E and H).

The measured populations of spherical/ellipsoidal vesi-cles and cylindrical/arachoidal vesicles at each time point(Table 3) were combined with additional unmeasured ves-icles (unmeasured vesicles were morphologically identifiedas spherical/ellipsoidal vesicles or cylindrical/arachoidalvesicles, but the images were poorly focused) into a groupcalled “observed” vesicles (Table 5). The frequencies ofoccurrence of measured and observed spherical/ellipsoidalvesicles and cylindrical/arachoidal vesicles at each timepoint were in close agreement (Table 5). A �2 analysis ofthe frequency of observed spherical/ellipsoidal vesicles andcylindrical/arachoidal vesicles at the five time points sug-gested a precursor-product relationship (Table 5): that thechanges in ratios of spherical/ellipsoidal vesicles and cylin-drical/arachoidal vesicles over time were significant.

DISCUSSION

Cryo-transmission electron microscopy is an excellent tech-nique for observing lipid-rich systems close to their original

FIGURE 5 Micrographs of fixed (OsO4) and negatively stained (1% NaPT) model bile sampled from an undisturbed tube at (a) 2 days and (B) 10 daysafter supersaturation by dilution. Note lamellar stacks (arrows) composed of discs of bilayer thickness 5.0–5.3 nm. Individual layers within the bilayersare visible (arrowheads). These lamellar structures were not seen in vitreous ice and are believed to be an artifact. Bar � 25 nm.

TABLE 3 Mean dimensions, surface area, and volume of spherical/ellipsoidal and cylindrical/arachoidal vesicles*

Time

Spherical/ellipsoidal vesicles Cylindrical/arachoidal vesicles

n Width Length Surface area Volume n Width Length Surface area Volume

6 h 124 54.9 � 6.7# 62.8 � 7.4##¶¶ 10,100 � 2,300# 102,000 � 37,000# 62 34.2 � 5.5*** 95.6 � 24.6§§§ 11,900 � 2,100§**** 86,000 � 20,000***

1 day 69 50.4 � 7.3§ 60.5 � 6.3§§¶¶ 8,700 � 2,100§ 82,000 � 27,000§ 141 29.3 � 5.2### 115.3 � 24.2** 11,800 � 2,400¶ ¶¶¶**** 77,000 � 24,000§§ ¶¶¶

4 days 122 55.8 � 8.2¶ 62.6 � 6.4¶ ¶¶ 10,400 � 2,700¶ 106,000 � 40,000¶ 133 27.3 � 2.8** 116.9 � 21.7¶ 11,200 � 2,100 69,000 � 17,000####

10 days 113 61.5 � 6.1** 65.8 � 6.0**¶¶ 12,300 � 2,300** 133,000 � 36,000** 75 27.8 � 3.1** 109.3 � 20.4** 10,700 � 1,700**** 66,000 � 14,000**####

20 days 114 55.1 � 4.3 59.6 � 4.6¶¶ 9,900 � 1,600 96,000 � 26,000 91 29.9 � 2.2 101.3 � 12.0 10,900 � 1,800**** 72,000 � 18,000####

*Data are expressed as mean � SD. Units: nm (width and length), nm2 (surface area), nm3 (volume). Surface area and volume data have been roundedoff. Results of statistical comparisons are based on Ansari-Bradley and Mann-Whitney tests.#p � 0.001 compared to 1 and 10 days. ***p � 0.001 compared to 1, 4, 10, and 20 days.§p � 0.001 compared to 4, 10, and 20 days. ###p � 0.05 compared to 4 and 10 days.¶p � 0.01 compared to 10 and 20 days. §§§p � 0.001 compared to 1, 4, and 10 days.**p � 0.01 compared to 20 days. ¶¶¶p � 0.05 compared to 4 days.##p � 0.05 compared to 1, 10, and 20 days. ****p � 0.001 compared to surface area of spherical/ellipsoidal vesicles.§§p � 0.001 compared to 10 days. ####p � 0.001 compared to volume of spherical/ellipsoidal vesicles¶¶p � 0.001 compared to width of spherical/ellipsoidal vesicles.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1445

Page 11: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

state, and our study of a supersaturated nucleating modelbile has provided unique observations regarding micelles,intermediate structures, and the evolution of small unila-mellar vesicles.

Micelle size and morphology

The size range of micelle width and length was 1.1–5.7 nmand 2.3–10 nm, respectively. These sizes are consistent withcoexisting simple and mixed micelles (Cohen and Carey,1990; Cohen et al., 1993; Mazer et al., 1980). The sizes ofsimple and mixed micelles in various model systems havebeen determined by techniques such as quasielastic lightscattering (Egelhaaf and Schurtenberger, 1994; Mazer et al.,1979), small-angle neutron scattering (Hjelm et al., 1992;Long et al., 1994b), and small-angle x-ray scattering(Somjen, 1994) and have previously been summarized (Co-hen et al., 1993; Gilat and Somjen, 1996).

The spherical/globular shape of simple micelles is widelyaccepted. The shape and structure of mixed micelles aresomewhat controversial and are strongly composition de-pendent (Cabral and Small, 1989). Several models for thestructure of bile salt-lecithin mixed micelles have beenproposed. The simple disc (Small, 1967) and mixed disc(Mazer et al., 1980) models are one bilayer thick and haveaxially oriented lecithin with exposed headgroups and a ringof bile salt molecules. Two models of elongated rodlike or“wormlike” micelles have been proposed. The stacked discmodel (Shankland, 1970) is composed of repeating units ofmixed discs with axially oriented lecithin molecules asmentioned above. A more popular model is an elongatedradial shell that has radially oriented lecithin molecules(originally proposed by Ulmius et al., 1982; modified to acapped rod model by Nichols and Ozarowski, 1990; sup-ported by Hjelm et al., 1992). Additional support for thecylindrical or “wormlike” mixed micelle model was pro-vided in egg yolk phosphatidylcholine (EYPC)/octyl glu-coside and EYPC/cholate systems by cryo-transmissionelectron microscopy (Vinson et al., 1989; Walter et al.,1991) and by quasielastic and static light scattering (Cohen

et al., 1998). The micelle population in our study appearedgenerally square to rectangular with occasional roundedends, suggesting a rodlike or barrel-like morphology. Inaddition, small, curled, threadlike structures of diameter1.1–2.2 nm were detected among micelles at 2 min aftersupersaturation (Fig. 3 A).

Intermediate structures

It is well documented that mixed micelles will undergotransition into small unilamellar vesicles upon critical dilu-tion of lecithin-bile salt mixtures (Cabral and Small, 1989;Mazer et al., 1980). The movement of bile salts into theintermicellar space to equalize bile salt concentration afterdilution destabilizes the micelles and promotes the forma-tion of intermediate structures such as patches of bilayer(Vinson et al., 1989; Walter et al., 1991), which could leadto the formation of small unilamellar vesicles.

The predominant intermediate structure in our systemwas denoted as a primordial vesicle. In a model bile ofcomposition similar to ours, low-contrast planar structureswere imaged by cryo-TEM at 1 min after dilution (Kaplunet al., 1997). These structures may be analogous to thecircular regions of reduced granularity that we observed at2 min after dilution (Fig. 3 A). Although the investigatorsspeculated that these planar structures could be lamellarcholesterol carriers (Somjen et al., 1990a) or intermediatestructures in a micelle-vesicle transition (Walter et al.,1991), no evidence was presented that temporally linked theplanar structures to small unilamellar vesicles. We believethat the primordial vesicle is that link.

One of the proposed mechanisms of vesicle assembly andformation is the gradual enlargement of disk-shaped mi-celles into circular sheetlike, bilayered phospholipid frag-ments of sufficient size and instability to close upon them-selves and form small unilamellar vesicles (Lasic, 1988).This enlargement into larger disks under depleted detergentconditions would require an edge minimization rate as highor higher than the rate of fusion of micelles to the enlargingdisk to maintain its smooth, round shape. A two-dimen-

TABLE 4 Mean surface area and volume of spherical/ellipsoidal and cylindrical/arachoidal vesicles*

Time n Surface area Volume Ratio#

6 h 186 10,700 � 2,400 96,000 � 33,000** 1:9.01 day 210 10,800 � 2,700 79,000 � 25,000## 1:7.34 days 255 10,800 � 2,400 86,000 � 35,000 1:8.010 days 188 11,600 � 2,200§ 106,000 � 44,000§§ 1:9.220 days 205 10,300 � 1,800¶ 86,000 � 26,000 1:8.3

*Measurements of spherical/ellipsoidal and cylindrical/arachoidal vesicles were combined at each time point, then averaged. Data are expressed as mean �SD. Units: nm2 (surface area), nm3 (volume). Surface area and volume data have been rounded off. Results of statistical comparisons are based onAnsari-Bradley and Mann-Whitney tests.#Surface area:volume.§p � 0.001 compared to 6 h and 1, 4, and 20 days.¶p � 0.01 compared to 6 h and 1 and 4 days.**p � 0.01 compared to 1, 4, 10, and 20 days.##p � 0.01 compared to 4, 10, and 20 days.§§p � 0.005 compared to 4 and 20 days.

1446 Biophysical Journal Volume 76 March 1999

Page 12: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

sional projection of this enlarging disk/sheet would show anintact bilayer around its entire perimeter.

In our model system the presence of cholesterol wouldincrease bilayer viscosity, reduce flexibility of the mem-brane, and greatly retard spontaneous closure of a roundunstable sheet. Therefore, we suggest the following mech-anism of vesicle assembly. Upon dilution and loss of bilesalts, unstable micelles could fuse at a rate higher than therate of edge minimization, resulting in a patchwork orframework of small portions of bilayer with many holesbetween. As this structure grows larger, its inherent flexi-bility (because it is not a solid sheet) allows a gradualincrease in curvature until a three-dimensional structure isformed. A two-dimensional projection of this holey struc-ture would appear as a generally circular primordial vesiclewith occasional well-formed bilayer segments such as thoseseen at 27 min after dilution (Fig. 3 C). Some primordial

vesicles viewed at this time point may be sheetlike becausethey were observed in thin ice toward hole centers (Fig. 3C), and three-dimensional morphology could not be con-firmed by 45° tilts. Other primordial vesicles were observedinterspersed among small unilamellar vesicles in thicker ice(Fig. 3 D). The micelles that appear to be trapped on theinside of young primordial vesicles (Fig. 3 B) (these mi-celles could be outside and be superimposed) as well asmicelles on the outside of primordial vesicles would con-tinue to merge with and eventually seal the surface of theforming vesicle. The growth trends of primordial vesicles(Fig. 6, A–C) and the diameters of small spherical/ellipsoi-dal vesicles at 6 h after dilution (Fig. 6 D) suggest a gradualenlargement. The larger primordial vesicles seen at 6 h afterdilution may produce large unilamellar vesicles, which wereseen at later time points (Table 1).

As noted above, primordial vesicles were observedface-on in thin ice most commonly away from hole edges.The extremely rare cluster of lamellae-like structures im-aged in Fig. 3 B (arrowhead) could be an artifact or a groupof sheetlike primordial vesicles viewed edge-on in thethicker ice. If these were edge views of primordial vesicles,we would expect to readily detect tangential views, but nonewere apparent.

Evolution of small unilamellar vesicles

Two distinct morphological populations of small unilamel-lar vesicles were identified in our system as spherical/ellipsoidal vesicles and cylindrical/arachoidal vesicles. Sev-eral investigators have observed small unilamellar vesiclesin human biles, as well as in other model biles, suggestingthat small unilamellar vesicles have a physiological role,serving as metastable cholesterol carriers (Cohen et al.,1989; Gilat and Somjen, 1996; Peled et al., 1988). In humanhepatic and gallbladder biles from patients with gallstones,spherical/ellipsoidal vesicles were observed using freeze

FIGURE 6 Size and morphological evolution of primordial vesicles (E),spherical/ellipsoidal vesicles (F), and cylindrical/arachoidal vesicles (�)with time. (A) 23 min, (B) 27 min, (C) 31 min, (D) 6 h, (E) 1 day, (F) 4days, (G) 10 days, (H) 20 days. The actual vesicle width is plotted versusthe ratio of width/length. Vesicles with the same dimensions are superim-posed. (A–D, F) Width/length ratios of primordial vesicles remained con-stant. The mean size of the population of primordial vesicles increasedsignificantly at 6 h and 4 days compared to 23, 27, and 31 min (Table 2).(D–H) The populations of spherical/ellipsoidal vesicles and cylindrical/arachoidal vesicles became more homogeneous with the passage of time(refer to Table 4 for mean sizes).

TABLE 5 Frequencies of observed and measured SEVsand CAVs*

Time

Observed SUVs# Measured SUVs§

SEVs CAVs SEVs CAVs

n % n % n % n %

6 hr 509¶ 64 288¶ 36 124 67 62 331 day 388¶ 38 626¶ 62 69 33 141 674 days 632¶ 51 611¶ 49 122 48 133 5210 days 487¶ 61 306¶ 39 113 60 75 4020 days 424¶ 49 436¶ 51 114 56 91 44Total 2440 52 2267 48 542 52 502 48

*SEVs, spherical/ellipsoidal vesicles; CAVs, cylindrical/arachoidal vesi-cles; SUVs, small unilamellar vesicles.#Observed SUVs included identifiable SEVs and CAVs in poorly focusedmicrographs and all measured SUVs. Note similarity of percentages tomeasured SUVs.§See Table 3.¶p � 0.01 for �2 analysis of frequencies at all five time points; �2 value �153, D.F. � 4.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1447

Page 13: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

fracture (Somjen et al., 1986). The composition of one ofthe gallbladder biles was 7.5% cholesterol, 18.8% phospho-lipid, and 73.7% bile salts (all mole %), a compositionsimilar to the model bile used in our study. Kaplun et al.(1994), using cryo-transmission electron microscopy, de-tected spherical/ellipsoidal vesicles in human gallbladderbile 1 day after collection and visualized spherical/ellipsoi-dal vesicles and arachoidal vesicles in human hepatic bile,known to contain cholesterol in vesicles (Pattison and Chap-man, 1986; Somjen et al., 1986), at 1, 3, 6, and 9 days aftercollection. In a model bile of composition 7.5% cholesterol,12.7% lecithin, and 79.7% sodium cholate (all mol%), Fu-dim-Levin et al. (1995) observed spherical and arachoidalvesicles by cryo-transmission electron microscopy 6 minafter mixing.

Although the initial goals of this study did not involvedetermining the exact formation time of spherical/ellipsoi-dal and cylindrical/arachoidal vesicles, the data stronglysuggest that spherical/ellipsoidal vesicles preceded cylindri-cal/arachoidal vesicles for the following reasons: 1) Primor-dial vesicles of the size and morphology seen 23, 27, and 31min after dilution (Fig. 3, B and C; Table 2) would likelybecome spherical/ellipsoidal vesicles (see 6-h time point inFig. 3 D and Table 3). 2) No primordial vesicles of cylin-drical/arachoidal shape were detected. 3) Only one cylin-drical/arachoidal vesicle (Fig. 3 C, at arrow) among 236spherical/ellipsoidal primordial vesicles was detected at 23,27, and 31 min (Fig. 3, B and C, Table 2) within 120photographed holes.

The mean surface area of the entire population of smallunilamellar vesicles was stable for 6 h through 4 days(Table 4), whereas the behavior of the population of spher-ical/ellipsoidal vesicles was more variable (Table 3). Theaggregation/fusion of a portion of the population of spher-ical/ellipsoidal vesicles with larger size to form multilamel-lar vesicles could explain the reduction in mean size andsurface area of spherical/ellipsoidal vesicles occurring at 1day (Table 3). However, the subsequent increase in meansize and surface area of the population of spherical/ellipsoi-dal vesicles noted at 4 days and 10 days (Table 3) whencholesterol monohydrate crystals had commenced to grow,presumably through the fused liquid crystal pathway (Wangand Carey, 1996a), is more difficult to reconcile. We spec-ulate that cholesterol and possibly lecithin was transferredfrom cylindrical/arachoidal vesicles to spherical/ellipsoidalvesicles (note reduction in surface area of cylindrical/ara-choidal vesicles at 4 and 10 days; Table 3). The surface areaof the population of spherical/ellipsoidal vesicles decreasedsignificantly at 20 days (Table 3), causing the mean surfacearea of the entire population of small unilamellar vesicles toreach a minimum (Table 4), whereas the surface area ofcylindrical/arachoidal vesicles was not significantlychanged (Table 3). This reduction in surface area of spher-ical/ellipsoidal vesicles likely reflects a major loss of cho-lesterol utilized in the production of arclike, helical, tubular,and cholesterol monohydrate crystals, which peaked at 18

days (Fig. 2), the latter originating in part from tubularcrystals (Wang and Carey, 1996a).

The process of aggregation/fusion of small unilamellarvesicles to form multilamellar vesicles that become visibleas liquid crystals by polarizing light microscopy is wellsupported (Halpern et al., 1986a,b; Kaplun et al., 1997). Inour polarizing light microscopy study, small liquid crystalswere initially observed at 1 day after supersaturation bydilution (Fig. 2) and disappeared after 14 days. Therefore,we expected to observe by cryo-transmission electron mi-croscopy aggregated and/or partially fused small unilamel-lar vesicles as well as many multilamellar vesicles. Theinability to detect aggregated or partially fused small unila-mellar vesicles might be explained in part by agitationbefore sampling, which could break aggregates, or by theirinherent low frequency. In any case, our study provides noinsight into how multilamellar structures are formed fromunilamellar structures. The multilamellar vesicles that wereseen at early time points (1 day in shaken bile, Table 1; 6 hin unshaken bile) were few and of small size (50–150 nm).Multilamellar vesicles as large as those observed by cryo-transmission electron microscopy at 20 days (Fig. 4 C) mayhave been artifactually enlarged by flattening in the icelayer and thus would not have been detectable by polarizinglight microscopy (Fig. 2), particularly because small andaggregated liquid crystals lack birefringence (Wang andCarey, 1996a,b). Liquid crystals or liquid crystal aggregateswith diameters of several microns would probably not becaptured in thin vitreous ice layers because the final samplevolume for cryo-transmission electron microscopy was�0.005 �l, whereas the sample volume for polarizing lightmicroscopy was 5 �l.

Not only the existence of multilamellar vesicles in bile invivo has been confirmed (Wang et al., 1997), but multila-mellar vesicles have been observed in human biles that weresubsequently incubated in vitro as well as in other modelbile systems (Reviewed in Cohen et al., 1993). Multilamel-lar vesicle-like particles were seen in vitrified samples ofhuman hepatic bile at 9 days after collection (Kaplun et al.,1994). Multilamellar vesicles were visualized by negativestain transmission electron microscopy (a technique proneto artifact with lipid systems) in gallbladder bile that washarvested from the interphase after a 90,000 � g centrifu-gation and incubated in vitro for 2 h (Halpern et al., 1986a).Multilamellar vesicles have also been observed in freeze-fractured preparations of model bile at 5 days of incubation(van de Heijning et al., 1994) and in vitreous ice prepara-tions of 48-h-old model bile (Gilat and Somjen, 1996).

Potential artifacts

This investigation provides qualitative and quantitative ob-servations of the size and shape of primordial and smallunilamellar vesicles in a model bile during a 20-day periodafter supersaturation. Potential morphological changes, suchas invaginations, could occur in vesicles when they are

1448 Biophysical Journal Volume 76 March 1999

Page 14: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

hyperosmotically stressed in increasing concentrations ofNaCl and bile salt during the blotting/previtrification pe-riod. Talmon (1996) has provided a useful discussion ofsome of the artifacts encountered with vitrification of com-plex fluids.

In our study, sample vitrification was performed in ahigh-humidity chamber, the use of which has been shown toprevent the formation of invaginations in lipid vesicles(Dubochet et al., 1988) and artifacts in surfactant disper-sions (Bellare et al., 1988). Our samples were vitrified onhydrophilic perforated support films, which are known toreduce the evaporation rate (Cyrklaff et al., 1990).

Despite the high-humidity environment surrounding thesample, progressive thinning of the surfactant film willoccur after blotting (Frederik et al., 1989). Water loss fromthe thinner area of the fluid film at the hole center in aperforated carbon film results in a biconcave-shaped film incross section, which promotes sorting of suspended parti-cles by size and shape (Frederik et al., 1989). Fig. 3 E showsan example of this sorting/ranking phenomenon. Biconcavefilm formation also promotes the concentration of vesiclesnear the hole edges. Thicker vitreous ice near hole edges didnot prevent the acquisition of vesicle images (i.e., Fig. 3, Dand E), enabling us to obtain a representative population ofsmall unilamellar vesicles.

As discussed in Materials and Methods, our pilot studyindicated that vitreous ice layers of optimum thickness wereobtained with postblotting/preplunging times of 1–10 s.During this period an increased concentration of NaCl couldhave hyperosmotically stressed vesicles, causing volumereduction and a resultant change in shape (i.e., ellipsoids tocylinders). However, a comparison of the frequencies ofeach morphological category of small unilamellar vesiclespresent in each of two aliquots of model bile withdrawn at1 day after supersaturation but vitrified after two differentblotting/preplunging times, 1 and 10 s, indicated that nomorphological changes had occurred. The ratios of spheri-cal and ellipsoidal vesicles to cylindrical and arachoidalvesicles were 1:2 at 1 s (48/94) and 10 s (22/46). Investi-gations by others into osmotic effects on vesicles havefocused primarily on the behavior of vesicles in hypotonicsolutions (Johnson and Buttress, 1973; Mui et al., 1993).

An increased concentration of taurocholate during theblotting/preplunging period could have promoted dissolu-tion of existing small unilamellar vesicles. Intermediatestructures such as bilayer sheets and cylindrical micelleshave been observed when phospholipid vesicles were ex-posed for at least 30 min to concentrations of cholate oroctyl glucoside near the micellar phase limit (Cohen et al.,1998; Vinson et al., 1989; Walter et al., 1991). We wouldnot be able to determine whether balanced dissolution ofspherical/ellipsoidal vesicles and cylindrical/arachoidal ves-icles occurred. However, the fact that the ratio of thesevesicle categories was stable at two different postblotting/preplunging times, 1 and 10 s, argues against unequaldissolution.

The relatively recent development of vitreous ice cryo-microscopy (Dubochet et al., 1988) has made it possible toimage lipid structures close to their original state in solution.Results obtained with vitreous ice-embedded lipid systemsmay now be compared to previous studies utilizing negativestaining (Howell et al., 1970; Somjen et al., 1990b), atechnique long believed to be prone to artifact in lipidsystems (reviewed by Cohen et al., 1993). We believe thatthe lamellar structures (Fig. 5) produced in negativelystained preparations of our model bile are artifacts, as hasbeen described in a recent review (Carey and Cohen, 1995).

SUMMARY

A supersaturated, nucleating model bile forming threephases (saturated micelles, liquid crystals, and cholesterolmonohydrate crystals) was examined by vitreous ice elec-tron microscopy. Micelles, primordial vesicles, two mor-phologically distinct populations of small unilamellar vesi-cles categorized as spherical/ellipsoidal vesicles andcylindrical/arachoidal vesicles, multilamellar vesicles, largeunilamellar vesicles, and cholesterol monohydrate crystalswere observed. Simple and mixed micelles had square/rectangular/ovoid/barrel-like morphology and mean widthand length of 3.6 and 5.6 nm. Intermediate structures be-tween micelles and unilamellar spherical/ellipsoidal vesi-cles were first seen at 23 min after supersaturation andrepresented the nucleation of primordial vesicles from su-persaturated micelles. Subtle changes in primordial vesiclestructure such as enlarged micelle-like particles in primor-dial vesicle interiors, faceted edges, and short segments ofbilayer at vesicle peripheries were observed over a timescale of minutes and suggested that vesicle formation in thismodel bile may be a gradual process involving the mergingof micelles to form vesicle boundaries. No evidence ofaggregation/fusion of small unilamellar vesicles to formmultilamellar vesicles was detected. The accepted role ofsmall unilamellar vesicles as metastable cholesterol carriersand nucleating agents was supported by the inverse corre-lation of a decrease in mean surface area of small unilamel-lar vesicles to an increase in production of cholesterolcrystals.

We thank Dr. Adrienne Cupples, Professor of Epidemiology and Biosta-tistics, Boston University School of Public Health, for assistance withstatistical analyses. We also thank Ms. Monika M. Leonard for experttechnical assistance.

Dr. Wang is a recipient of an Industry Research Scholar Award from theAmerican Digestive Health Foundation/American Gastroenterological As-sociation (1996–1999). This work was supported in part by research andcenter grants DK 34584 (DQ-HW), DK52911, DK 36588, and DK 34854(MCC), and HL 26335–17 (DMS), all from the National Institutes ofHealth (U.S. Public Health Service).

REFERENCES

Bartlett, G. R. 1959. Phosphorous assay in column chromatography.J. Biol. Chem. 234:466–468.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1449

Page 15: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

Bellare, J. R., H. T. Davis, L. E. Scriven, and Y. Talmon. 1988. Controlledenvironment vitrification system: an improved sample preparation tech-nique. J. Electron Microsc. Tech. 10:87–111.

Cabral, D., and D. M. Small. 1989. Physical chemistry of bile. In Hand-book of Physiology: The Gastrointestinal System III. S. G. Schultz, J. G.Forte, and B. B. Rauner, editors. American Physiological Society, Wa-verly Press, Baltimore. 621–662.

Carey, M. C. 1978. Critical tables for calculating the cholesterol saturationof native bile. J. Lipid Res. 19:945–955.

Carey, M. C., and D. E. Cohen. 1995. Update on physical state of bile. Ital.J. Gastroenterol. 27:92–100.

Carey, M. C., and D. M. Small. 1978. The physical chemistry of choles-terol solubility in bile: relationship to gallstone formation and dissolu-tion in man. J. Clin. Invest. 61:998–1026.

Cohen, D. E., M. Angelico, and M. C. Carey. 1989. Quasielastic lightscattering evidence for vesicular secretion of biliary lipids.Am. J. Physiol. 257:G1–G8.

Cohen, D. E., and M. C. Carey. 1990. Rapid 1 hour high-performance gelfiltration chromatography resolves coexisting simple micelles, mixedmicelles and vesicles in bile. J. Lipid Res. 31:2103–2112.

Cohen, D. E., M. R. Fisch, and M. C. Carey. 1990. Principles of laserlight-scattering spectroscopy: applications to the physicochemical studyof model and native biles. Hepatology. 12:113S–122S.

Cohen, D. E., E. W. Kaler, and M. C. Carey. 1993. Cholesterol carriers inhuman bile: are “lamellae” involved? Hepatology. 18:1522–1532.

Cohen, D. E., G. M. Thurston, R. A. Chamberlin, G. B. Benedek, and M. C.Carey. 1998. Laser light scattering evidence for a common wormlikegrowth structure of mixed micelles in bile salt- and straight-chaindetergent-phosphatidylcholine aqueous systems: relevance to the micel-lar structure of bile. Biochemistry. 37:14798–14814.

Cyrklaff, M., M. Adrian, and J. Dubochet. 1990. Evaporation duringpreparation of unsupported thin vitrified aqueous layers for cryo-electron microscopy. J. Electron Microsc. Tech. 16:351–355.

Dubochet, J., M. Adrian, J. J. Chang, J. C. Homo, J. Lepault, A. W.McDowell, and P. Schultz. 1988. Cryo-electron microscopy of vitrifiedspecimens. Q. Rev. Biophys. 21:129–228.

Dubochet, J., M. Groom, and S. Mueller-Neuteboom. 1982. The mountingof macromolecules for electron microscopy with particular reference tosurface phenomena and the treatment of support films by glow dis-charge. Adv. Opt. Electron Microsc. 8:107–135.

Egelhaaf, S. V., and P. Schurtenberger. 1994. Shape transformation in thelecithin-bile salt system: from cylinders to vesicles. J. Phys. Chem.98:8560–8573.

Frame, J. S. 1992. Ellipsoid and Spheroid. In Encyclopedia of Science andTechnology, Vol. 6. McGraw-Hill, New York. 315–316.

Frederik, P. M., M. C. A. Stuart, P. H. H. Bomans, and W. M. Busing.1989. Phospholipid, nature’s own slide and cover slip for cryo-electronmicroscopy. J. Microsc. 153:81–92.

Fromm, H., P. Amin, H. Klein, and I. Kupke. 1980. Use of a simpleenzymatic assay for cholesterol analysis in human bile. J. Lipid Res.21:259–261.

Fudim-Levin, E., A. Bor, A. Kaplun, Y. Talmon, and D. Lichtenberg.1995. Cholesterol precipitation from cholesterol-supersaturated bilemodels. Biochim. Biophys. Acta. 1259:23–28.

Gilat, T., and G. J. Somjen. 1996. Phospholipid vesicles and other choles-terol carriers in bile. Biochim. Biophys. Acta. 1286:95–115.

Groen, A. K., R. Ottenhoff, P. L. M. Jansen, J. van Marle, and G. N. J.Tytgat. 1989. Effect of cholesterol nucleation-promoting activity oncholesterol solubilization in model bile. J. Lipid Res. 30:51–58.

Halpern, Z., M. A. Dudley, A. Kibe, M. P. Lynn, A. C. Breuer, and R. T.Holzbach. 1986a. Rapid vesicle formation and aggregation in abnormalhuman biles. A time-lapse video-enhanced contrast microscopy study.Gastroenterology. 90:875–885.

Halpern, Z., M. A. Dudley, M. P. Lynn, J. M. Nader, A. C. Breuer, andR. T. Holzbach. 1986b. Vesicle aggregation in model systems of super-saturated bile: relation to crystal nucleation and lipid composition of thevesicular phase. J. Lipid Res. 27:295–306.

Hjelm, R. P., P. Thiyagarajan, and H. Alkan-Onyuksel. 1992. Organizationof phosphatidylcholine and bile salt in rodlike mixed micelles. J. Phys.Chem. 96:863–866.

Howell, J. I., J. A. Lucy, R. C. Pirola, and I. A. D. Bouchier. 1970.Macromolecular assemblies of lipid in bile. Biochim. Biophys. Acta.210:1–6.

Johnson, S. M., and N. Buttress. 1973. The osmotic insensitivity ofsonicated liposomes and the density of phospholipid-cholesterol mix-tures. Biochim. Biophys. Acta. 307:20–26.

Kaplun, A., F. M. Konikoff, A. Eitan, M. Rubin, A. Vilan, D. Lichtenberg,T. Gilat, and Y. Talmon. 1997. Imaging supramolecular aggregates inbile models and human bile. Microsc. Res. Tech. 39:85–96.

Kaplun, A., Y. Talmon, F. M. Konikoff, M. Rubin, A. Eitan, M. Tadmor,and D. Lichtenberg. 1994. Direct visualization of lipid aggregates innative human bile by light- and cryo-transmission electron microscopy.FEBS Lett. 340:78–82.

Klosgen, B., and W. Helfrich. 1993. Special features of phosphatidylcho-line vesicles as seen in cryo-transmission electron microscopy. Eur.Biophys. J. 2:329–340.

Konikoff, F. M., and M. C. Carey. 1994. Cholesterol crystallization froma dilute bile salt-rich model bile. J. Cryst. Growth. 144:79–86.

Konikoff, F. M., D. S. Chung, J. M. Donovan, D. M. Small, and M. C.Carey. 1992. Filamentous, helical, and tubular microstructures duringcholesterol crystallization from bile. J. Clin. Invest. 90:1155–1160.

Lasic, D. D. 1988. The mechanism of vesicle formation. Biochem. J.256:1–11.

Little, T. E., S. P. Lee, H. Madani, E. W. Kaler, and K. Chinn. 1991.Interconversions of lipid aggregates in rat and model bile.Am. J. Physiol. 260:G70–G79.

Long, M. A., E. W. Kaler, and S. P. Lee. 1994a. Structural characterizationof the micelle-vesicle transition in lecithin-bile salt solutions. Biophys. J.67:1733–1742.

Long, M. A., E. W. Kaler, S. P. Lee, and G. D. Wignall. 1994b. Charac-terization of lecithin-taurodeoxycholate mixed micelles using small-angle neutron scattering and static and dynamic light scattering. J. Phys.Chem. 98:4402–4410.

Mazer, N. A., G. B. Benedek, and M. C. Carey. 1980. Quasielasticlight-scattering studies of aqueous biliary lipid systems. Mixed micelleformation in bile salt-lecithin solutions. Biochemistry. 19:601–615.

Mazer, N. A., and M. C. Carey. 1983. Quasi-elastic light-scattering studiesof aqueous biliary lipid systems. Cholesterol solubilization and precip-itation in model bile solutions. Biochemistry. 22:426–442.

Mazer, N. A., M. C. Carey, R. F. Kwasnick, and G. B. Benedek. 1979.Quasielastic light scattering studies of aqueous biliary lipid systems.Size, shape, and thermodynamics of bile salt micelles. Biochemistry.18:3064–3075.

Mui, B. L.-S., P. R. Cullis, E. A. Evans, and T. D. Madden. 1993. Osmoticproperties of large unilamellar vesicles prepared by extrusion. Bio-phys. J. 64:443–453.

Mui, B. L.-S., H.-G. Dobereiner, T. D. Madden, and P. R. Cullis. 1995.Influence of transbilayer area asymmetry on the morphology of largeunilamellar vesicles. Biophys. J. 69:930–941.

Nichols, J. W., and J. Ozarowski. 1990. Sizing of lecithin-bile salt mixedmicelles by size-exclusion high-performance liquid chromatography.Biochemistry. 29:4600–4606.

Pattison, N., and B. Chapman. 1986. Distribution of biliary cholesterolbetween mixed micelles and non-micelles in relation to fasting andfeeding in humans. Gastroenterology. 91:697–702.

Peled, Y., Z. Halpern, R. Baruch, G. Goldman, and T. Gilat. 1988.Cholesterol nucleation from its carriers in human bile. Hepatology.8:914–918.

Pope, J. L. 1967. Crystallization of sodium taurocholate. J. Lipid Res.8:146–147.

Shankland, W. 1970. The equilibrium and structure of lecithin-cholatemixed micelles. Chem. Phys. Lipids. 4:109–130.

Small, D. M. 1967. Physico-chemical studies of cholesterol gallstoneformation. Gastroenterology. 52:607–610.

Small, D. M. 1980. Cholesterol nucleation and growth in gallstone forma-tion. N. Engl. J. Med. 302:1305–1307.

1450 Biophysical Journal Volume 76 March 1999

Page 16: Cryoelectron Microscopy of a Nucleating Model Bile in ... · Cryoelectron Microscopy of a Nucleating Model Bile in Vitreous Ice: Formation of Primordial Vesicles Donald L. Gantz,*

Small, D. M. 1986. Sterols and Sterol Esters. In The Physical Chemistry ofLipids from Alkanes to Phospholipids. Plenum Press, New York.395–406.

Somjen, G. 1994. New concepts in bile salt/phospholipid systems. InTheory and Applications. Swiss Society for Biochemistry, Swiss Bio-physical Society, Swiss Group of Colloid and Interface Science andHoffmann-LaRoche International Symposium, Basel, May.

Somjen, G. J., Y. Marikovsky, P. Lelkes, and T. Gilat. 1986. Cholesterol-phospholipid vesicles in human bile: an ultrastructural study. Biochim.Biophys. Acta. 879:14–21.

Somjen, G. J., Y. Marikovsky, E. Wachtel, P. R. C. Harvey, R. Rosenberg,S. M. Strasberg, and T. Gilat. 1990a. Phospholipid lamellae are choles-terol carriers in human bile. Biochim. Biophys. Acta. 1042:28–35.

Somjen, G. J., R. Rosenberg, and T. Gilat. 1990b. Gel filtration andquasielastic light scattering studies of human bile. Hepatology. 12:123S–129S.

Talmon, Y. 1996. Transmission electron microscopy of complex fluids: thestate of the art. Ber. Bunsenges. Phys. Chem. 100:364–372.

Tao, S., S. Tazuma, and G. Kajiyana. 1993. Apolipoprotein A-I stabilizesphospholipid lamellae and thus prolongs nucleation time in model bilesystems: an ultrastructural study. Biochim. Biophys. Acta. 1166:25–30.

Toor, E. W., D. F. Evans, and E. L. Cussler. 1978. Cholesterol monohy-drate growth in model bile solutions. Proc. Natl. Acad. Sci. USA.75:6230–6234.

Toyoshima, C. 1989. On the use of holey grids in electron crystallography.Ultramicroscopy. 30:439–444.

Turley, S. D., and J. M. Dietschy. 1978. Re-evaluation of the 3�-hydroxysteroid dehydrogenase assay for total bile acids in bile. J. LipidRes. 19:924–928.

Ulmius, J., G. Lindblom, H. Wennerstrom, L. B.-A. Johansson, K. Fontell,O. Soderman, and G. Arvidson. 1982. Molecular organization in theliquid-crystalline phases of lecithin-sodium cholate-water systems stud-ied by nuclear magnetic resonance. Biochemistry. 21:1553–1560.

van de Heijning, B. J. M., M. F. J. Stolk, K. J. van Erpecum, W. Renooij,and G. P. van Berge Henegouwen. 1994. The effects of bile salt hydro-phobicity on model bile vesicle morphology. Biochim. Biophys. Acta.1212:203–210.

Vinson, P. K., Y. Talmon, and A. Walter. 1989. Vesicle-micelle transitionof phosphatidylcholine and octyl glucoside elucidated by cryo-transmission electron microscopy. Biophys. J. 56:669–681.

Walter, A., P. K. Vinson, A. Kaplun, and Y. Talmon. 1991. Intermediatestructures in the cholate-phosphatidylcholine vesicle-micelle transition.Biophys. J. 60:1315–1325.

Wang, D. Q-H., and M. C. Carey. 1996a. Complete mapping of crystalli-zation pathways during cholesterol precipitation from model bile: influ-ence of physical-chemical variables of pathophysiologic relevance andidentification of a stable liquid crystalline state in cold, dilute andhydrophilic bile salt-containing systems. J. Lipid Res. 37:606–630.

Wang, D. Q-H., and M. C. Carey. 1996b. Characterization of crystalliza-tion pathways during cholesterol precipitation from human gallbladderbiles: identical pathways to corresponding model biles with three pre-dominating sequences. J. Lipid Res. 37:2539–2549.

Wang, D. Q-H., B. Paigen, and M. C. Carey. 1997. Phenotypic character-ization of Lith genes that determine susceptibility to cholesterol chole-lithiasis in inbred mice: physical chemistry of gallbladder bile. J. LipidRes. 38:1395–1411.

Gantz et al. Cryomicroscopy of Nucleating Model Bile 1451